ww.sciencedirect.com

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6

Available online at w

ScienceDirect

journal homepage: www.elsevier .com/locate /watres

Removal of pharmaceuticals in aerated biofilterswith manganese feeding

Yongjun Zhang a,*, Hong Zhu b, Ulrich Szewzyk b, Sven Uwe Geissen a

a Technische Universit€at Berlin, Chair of Environmental Process Engineering, Sekr. KF2, Strasse des 17.Juni 135,

10623 Berlin, Germanyb Technische Universit€at Berlin, Chair of Environmental Microbiology, Sekr. BH 6-1, Ernst-Reuter-Platz 1, 10587

Berlin, Germany

a r t i c l e i n f o

Article history:

Received 31 May 2014

Received in revised form

16 December 2014

Accepted 7 January 2015

Available online 14 January 2015

Keywords:

Biodegradation

Biofiltration

Tertiary treatment

Emerging pollutant

Biofilm

* Corresponding author. Tel.: þ49 (0)30 31425E-mail address: yongjun.zhang@tu-berlin

http://dx.doi.org/10.1016/j.watres.2015.01.0090043-1354/© 2015 Elsevier Ltd. All rights rese

a b s t r a c t

A tertiary treatment step is required in current wastewater treatment plants to remove

trace pollutants and thus to prevent their extensive occurrence in the aquatic environ-

ment. In this study, natural MnOx ore and natural zeolite were separately used to pack two

lab-scale aerated biofilters, which were operated in approximately 1.5 years for the

removal of frequently occurring pharmaceuticals, including carbamazepine (CBZ), diclo-

fenac (DFC), and sulfamethoxazole (SMX), out of synthetic and real secondary effluents.

Mn2þ was added in the feeds to promote the growth of iron/manganese oxidizing bacteria

which were recently found to be capable of degrading recalcitrant pollutants. An effective

removal (80e90%) of DFC and SMX was observed in both biofilters after adaptation while a

significant removal of CBZ was not found. Both biofilters also achieved an effective removal

of spiked Mn2þ, but a limited removal of carbon and nitrogen contents. Additionally, MnOx

biofilter removed 50% of UV254 from real secondary effluent, indicating a high potential on

the removal of aromatic compounds.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Various pharmaceutical active compounds (PhACs) have been

detected in water bodies. Although at trace concentrations,

they still have a potential impact on the aquatic ecological

system (Brausch et al., 2012). This impact could become

complex within the interactions of various species. For

instance, the antibiotics would lead to an increased resistance

ofmicroorganisms who can spread their resistance genes into

other microbes (Baquero et al., 2008). It was also reported that

the dramatic population decline of vultures in the India sub-

continent was resulted from the residue of diclofenac (DFC) in

298; fax: þ49 (0)30 31425.de (Y. Zhang).

rved.

livestock bodies which held an important position in the food

chain of vultures (Green et al., 2004). Main sources of PhACs

are the human and veterinary applications. PhACs from the

former source would mostly enter wastewater treatment

plants (WWTPs) via a sewage system while those from the

latter are normally diffused. Therefore, one important strat-

egy to control the contamination of PhACs is to upgrade the

current WWTPs which currently cannot effectively eliminate

all PhACs (Zhang and Geißen, 2010).

Many technologies have been studied for the removal of

PhACs, including AOP, ozonation, membrane separation,

sorption, etc (Basile et al., 2011). Among them, ozonation and

activated carbon adsorption are most promising and have

487.

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6 219

been installed at full scale (Reungoat et al., 2010;

Zimmermann et al., 2011). However, those chemical/physical

methods mostly require high investment and/or operation

costs. Therefore, the development of a cost-efficient method

is of great concern.

Iron/manganese oxidizing bacteria (IMOB) can mediate the

oxidation of Fe2þ and Mn2þ to their high-valency states and

thus play an important role in the natural cycle of iron and

manganese (Emerson et al., 2010). IMOBwidely exist in natural

water bodies and technical water systems. For example,

Cerrato et al. (2010) detected several strains of IMOB in the

sediment of surface water, on the filter materials of drinking

water treatment, and in thewater distribution pipes evenwith

the presence of chlorine. IMOB have been widely involved in

the removal of Fe2þ and Mn2þin the drinking water treatment

(Gouzinis et al., 1998).

Recently, researchers found that IMOB are also capable of

degrading PhACs. One degradation mechanism is the in vitro

adsorption and oxidation by the biogenic manganese oxides

(Bio-MnOx). Sabirova et al. (2008) studied the removal of 17a-

ethinylestradiol (EE2) with several IMOB strains and they

found that IMOB can effectively degrade EE2with the presence

of Mn2þ. When a bacteriostatic agent (sodium azide) was

applied to inhibit the manganese-oxidizing activity of IMOB,

the authors still observed 80% removal of EE2. Thus they

concluded that the degradation was initiated by the attack of

biogenic manganese oxides (Bio-MnOx). This is in accordance

with the fact that manganese oxides hold a high redox po-

tential and both synthetic and biogenic MnOx are able to

oxidize many organic and inorganic pollutants (Hennebel

et al., 2009; Zaman et al., 2009; Zhang et al., 2008a). However,

it was also found that the participation of active IMOB was

beneficial and sometimes essential for pollutant oxidation.

Murray and Tebo (2007) found that IMOB Bacillus sp. strain SG-

1 can accelerate the oxidation of Cr3þ compared to both syn-

thetic and biogenicMnOx. Forrez et al. (2010) reported that the

addition of either sodium azide or lysozyme (a cell lysis agent)

resulted in a significant inhibition of DFC oxidation by Bio-

MnOx of IMOB Pseudomonas putida. Meerburg et al. (2012)

further reported that the heat inactivated Bio-MnOx of P.

putida cannot remove DFC. Our previous study also observed

that several IMOB strains can remove DFC but the heat inac-

tivated biomass resulted in no removal (Zhu et al., 2012).

Therefore, other mechanisms may exit to initiate the

pollutant removal which occurs with the active manganese-

oxidizing metabolism of IMOB, e.g. via the involvement of a

ligand-bound Mn3þ intermediate (Meerburg et al., 2012).

Therefore, for real applications, the active involvement of

IMOB should be guaranteed.

However, only few of studies have been published on the

removal PhACs with IMOB in technical systems. De Rudder

et al. (2004) used a MnO2 mineral as a packing material in a

filter for the removal of EE2 from tap water. They noticed that

the loaded EE2 on MnO2 was significantly beyond its adsorp-

tion capacity and thus they proposed that a self-regeneration

process may occur. Forrez et al. (2009) compared two filters

packed with the same mineral and plastic polyethylene rings.

The filter with plastic rings was additionally fed with Mn2þ

and therefore the rings were expected to be covered with Bio-

MnOx. Both filters can successfully remove EE2 but the

biofilter with plastic rings failed to recover the effective

removal after an increase of EE2 loading and also resulted in a

higher Mn2þ level in effluent than the filter with MnO2 min-

eral. Recently, Forrez et al. (2011) added Bio-MnOx

(160 mg Mn4þ L�1) into the membrane module of a biore-

actor and achieved an effective removal of many PhACs.

However, such a configuration is complex to operate and may

have some negative impacts on the membrane performance,

e.g. fouling, permeability, etc.

In the present study, aerated biofilters were studied, with

natural MnOx ore and zeolite as packing materials. Mn2þ was

additionally spiked into the feeding wastewater to promote

the growth of IMOB. Therefore, a simple configuration and

active IMOB were expected with the studied aerated biofilters

and the removal of three widely detected PhACs, including

DFC, carbamazepine (CBZ) and sulfamethoxazole (SMX), was

investigated with both synthetic and real wastewater.

2. Material and methods

2.1. Materials

MnOx ore (Aqua-mandix®) and natural zeolite (Bigadia®) were

supplied by Aqua-Techniek, Netherlands, and EgeZeolit,

Turkey, respectively. Their chemical compositions and phys-

ical properties are shown in Table 1. Diclofenac sodium, car-

bamazepine, sulfamethoxazole (all with analytical grade), and

manganese(II) sulfate monohydrate with ReagentPlus® grade

were purchased from Sigma Aldrich, Germany.

2.2. Biofilter and operation

The MnOx ore and zeolite were used to pack two biofilters,

named Mn-Biofilter and Zeo-Biofilter, respectively. Both bio-

filters were constructed out of cylindrical acrylic glass

(diameter 7 cm; effective height 45 cm; packed height 35 cm)

andwere aeratedwith a fine-bubblemembrane diffuser at the

bottom (Fig. 1). Aluminum foil was used to wrap the filters to

prevent light. The aeration rate was regulated with a needle

valve at approximately 200 mL h�1 and 3e5 mg L�1 dissolved

oxygen was achieved in both biofilters. The biofilters were fed

with wastewater (compositions described below) via a multi-

channel peristaltic pump at 430 mL d�1, which resulted in

an empty bed contacting time (EBCT) of 28 h and 20 h in Mn-

Biofilter and Zeo-Biofilters, respectively. In order to avoid the

clogging problem, another multi-channel peristaltic pump

was used to recycle water in both biofilters at 300 mL h�1 to

generate a superficial velocity of 3 and 2 m h�1 in Mn-Biofilter

and Zeo-Biofilters, respectively.

Both synthetic wastewater and effluent from a localWWTP

were applied to feed the biofilters. The synthetic wastewater

was produced via diluting the OECD composition (OECD, 2001)

to simulate a WWTP effluent: 1-L tap water containing

peptone 16mg,meat extract 11mg, NH4Cl 5mg, NaNO3 10mg,

K2HPO4 2.8 mg, NaCl 0.7 mg, CaCl2$2H2O 0.4 mg, Mg2SO4$7H2O

0.2 mg (DOC 10 mg L�1, TN 9 mg L�1, pH 7.5). The collected

WWTP effluent containedDOC 11 ± 3mg L�1, TN 15 ± 2mg L�1,

TP 0.6 mg L�1, conductivity 1260 ± 50 mS cm�1, UV254 0.3 cm�1,

pH 7.0. In addition, 20 mg L�1 Mn2þ was added into both

Table 1 e Chemical composition and physical properties of MnO2 and zeolite used in the biofilters.

MnO2 Composition MnO2 78%, Fe2O3 6.2%, SiO2 5.2%, Al2O3 3.1%

Size 0.5e1.0 mm

Bulk density 2000 kg m�3

Void ratio 0.3

Zeolite Composition SiO2 70.14%, Al2O3 11.46%, K2O 4.37%, CaO 2.98%, MgO 1.46%, Fe2O3 0.85%

Size 2e5 mm

Bulk density 1000 kg m�3

Void ratio 0.4

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6220

feedings. SMX, CBZ and DFCwere also spiked in both feedings

at concentrations described below.

The biofilters were initially filled with a solution of 10-d old

IMOB strains (Leptothrix sp. and Pseudomonas sp.) which were

separately cultivated in a liquid medium (lepthothrix strains

media 2) prepared according to (Atlas, 1997). After 2 d, the

biofilters were continuously fed with the synthetic

wastewater.

Both biofilters experienced the following operation phases:

A) starting with 2 mg L�1 PhACs in synthetic wastewater; B)

working phase with the same feeding; C) feeding with

1.5 mg L�1 PhACs in synthetic wastewater; D) feeding with

1.5 mg L�1 extra PhACs in WWTP effluent; E) feeding with

1.5 mg L�1 extra PhACs in WWTP effluent without feeding

Mn2þ. The final concentrations of SMX, CBZ andDFC inWWTP

effluent were 2, 5, 4 mg L�1, respectively. The biofilters were

assumed to enter the working phase (Phase B) when an

effective SMX removal was observed. A transition period of

2e4 weeks occurred between the other sequent operation

phases.

3 samples were taken per week in the operation Phase A

and B and 2 samples in the other phases. After filtration with

0.45 mm cellulose nitrate membrane, all samples were stored

at 4 �C for the collective analysis.

2.3. Control tests

Some carrier materials were taken from both biofilters and a

defined amount (wet weight: 10 g Mn ore, 20 g zeolite) was

Fig. 1 e The scheme of an aerated biofilter.

added into 50 mLWWTP effluent spiked with 2 mg L�1 PhACs.

A control test with the presence of sodium azide (50 mg L�1)

was also conducted for bothmaterials. Samples of water were

taken in intervals during 10 d and analyzed with HPLC after

centrifugation.

2.4. Sample analysis

PhACs at the high concentrationwere analyzedwith HPLC-UV

(Agilent 1200) system equippedwith a Gemini-NX C18 column

(250 � 4.6 mm, 5 mm) and a SecurityGuard pre-column. The

column temperature was set at 30 �C. The injection volume

was 50 mL. Elution was conducted with Millipore water (0.1%

formic acid) and acetonitrile, at a flow rate of 1.0 mL min�1

with the following gradient: 40% acetonitrile at 0 min, 60% at

8 min, 90% at 9 min, 60% at 14 min, and 40% at 15 min (the

end). The eluted SMX, CBZ and DFC were detected by a UV

detector at 265 nm, 285 nm, and 275 nm, respectively. The

limit of quantification was 50 mg L�1 for all three PhACs.

The samples with low concentrations of PhACs were

analyzed with a HPLC-MS system, consisting of Agilent 1260

HPLC and AB SciexQTrap 5500 MS with an electrospray ioni-

zation interface. Analysis was carried out in positive ion

mode. A Synergi Hydro-RP column (150 � 3 mm) with a Se-

curity Guard column AQ C18 was used for the chromato-

graphic separation. 70 mL of sample was injected into the

HPLC-MS system. The column oven temperature was set to

40 �C. The elution solution consisted of Millipore water with

0.1% formic acid and 10 m Mammonium format (A) and

methanol with 0.1% formic acid (B). The gradient elution

program was set as follows: 0e4 min, 98% (A); 5 min, 60%;

16.5 min, 0%; 22 min, 0%; 22.1 min, 98%. The run time was

28 min and the flow rate was 0.50 mL min�1.

Dissolved organic carbon (DOC) and total Nitrogen (TN)

were analyzedwith aDOC/TN analyzer (Analytik Jenamulti N/

C 3100). DOC was measured as non-purgeable organic carbon.

Mn2þ was analyzed by an atomic absorption spectroscopy

(PerkinElmer PinAAcle 900 Z). The absorbance of ultraviolet at

254 nm (UV254) was measured with a photo spectrometer

(DR5000, Hach Lange GmbH). Total phosphorus (TP) was

measured with a cuvette test from Hach Lange GmbH.

3. Results

3.1. Removal of PhACs

Fig. 2a shows the removal efficiencies of SMX, CBZ, and DFC in

Mn-Biofilter at different operation phases. About 95% of DFC

Fig. 3 e UVevis spectra of SMX and DFC in Mn-Biofilter (a)

and Zeo-Biofilter (b).

Fig. 2 e Removal efficiencies of SMX, CBZ and DFC in Mn-

Biofilter (a) and Zeo-Biofilter (b) in different operation

phases: A) starting with 2 mg L¡1 PhACs in synthetic

wastewater; B) working phase with the same feeding; C)

feeding with 1.5 mg L¡1 PhACs in synthetic wastewater; D)

feeding with 1.5 mg L¡1 extra PhACs in WWTP effluent; E)

feeding with 1.5 mg L¡1 extra PhACs in WWTP effluent

without feeding Mn2þ. The whiskers show 5% and 95%

percentiles and the solid dots stand for the mean value.

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6 221

was removed immediately after starting the biofilter. On the

contrary, only 10% of SMX was removed in the starting period

of 3 months. However, thereafter the removal of SMX rapidly

increased to >90% in several days (defined as the start of

operation Phase B) and kept stable. The removal of both SMX

and DFC was dropped to 70% and 80% (median value),

respectively, after switching to the trace concentration (Phase

C). More interestingly, both removal efficiencies were elevated

with WWTP effluent (Phase D): 90% of SMX and 96% of DFC.

After excluding Mn2þ in the feeding (Phase E), the removal of

both SMX and DFC was slightly decreased and a broader

variation was also observed here than in the previous phase.

CBZ obtained a negative removal efficiency in Mn-Biofilter

after the trace feeding concentrations were applied, which

indicates that the removal of CBZ in Phase A and B, although

at very low efficiencies, was probably due to the sorption onto

the carrier material.

The removal of SMX, CBZ, and DFC in Zeo-Biofilter is

shown in Fig. 2b. It can be found that all three substanceswere

poorly removed in the approximately 4-month starting phase.

However, an effective removal of SMX appeared quickly after

the starting phase, just like the Mn-Biofilter. An escalation of

DFC removal was clearly observed through the operation

phases and finally a median removal of 88% was achieved in

phase D after about 1-year operation. A slightly decreased

removal of SMX (79%) and DFC (74%) was also found in Phase

E, similar to Mn-Biofilter. This biofilter also resulted in a

limited removal of CBZ in Phases A-D but it was very inter-

esting that CBZ was slightly removed (<20%) in Phase E.

UVevis spectroscopy was applied to evaluate the evolution

of aromatic contents in the treated wastewater. At the end of

Phase E, 10 mg L�1 DFC or SMX was separately fed into both

biofilters and the effluents were scanned with UVevis spec-

troscopy. The UVevis spectra of SMX and DFC are shown in

Fig. 3. It can be found that Mn-Biofilter effectively destroyed

the aromatic structure of DFC while SMX was only partially

removed (Fig. 3a). An opposite result was observed in Zeo-

Biofilter where the aromatic structure of SMX was

completely eliminated (Fig. 3b). Nevertheless, the partial

removal of DFC or SMX may be due to the high loading, it

might be possible that the aromatic structure of both

Fig. 4 e Concentration profiles of DFC and SMX

(C0 ¼ 2mg L¡1) in wastewater andwastewater plus sodium

azide with the inoculated carrier materials from Mn-

Biofilter (a) and Zeo-Biofilter (b). Error bars stand for the

standard deviation of triple tests.

Fig. 5 e Effluent (C) to influent (C0) ratios of DOC and TN in

Mn-Biofilter (a) and Zeo-Biofilter (b) with synthetic

wastewater and WWTP effluent as feedings. The whiskers

show 5% and 95% percentiles and the solid dots stand for

the mean value.

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6222

substances can also be destroyed at a trace level. However, the

transformation products were not identified in this study,

which makes the degradation pathways still unclear.

A control test was conducted in order to clarify the bio-

logical and chemical effects of the biomass on both carrier

materials. As shown in Fig. 4, the presence of sodium azide

significantly depressed the elimination of SMX by both carrier

materials, indicating that the elimination of SMX can be

attributed to the biological oxidation of manganese in both

biofilters. Azide also negatively influenced DFC removal with

the inoculated zeolite but its influence on DFC removal with

inoculated MnOx ore was limited, which indicates a signifi-

cant removal by chemical reactions.

3.2. Removal of DOC, TN, TP, and UV254

DOC and TN were monitored during the whole operation of

biofilters while TP and UV254 only when WWTP effluent was

applied. Fig. 5 shows the C/C0 ratios of DOC and TN in three

phases: feeding with synthetic wastewater plus Mn, with

WWTP effluent plusMn, andwithWWTP effluentwithoutMn.

The absolute concentrations of DOC and TN can also been

found in Supplementary Information (Fig. 1s). In general, both

biofilters resulted in a limited removal of DOC and TN. DOC

removal was slightly enhanced in both biofilters when they

were fed with WWTP effluent plus Mn: Median values of C/C0

were deceased from 83% to 64% and 92%e76% in Mn-Biofilter

and Zeo-Biofilter, respectively. However, when Mn was

excluded in the feed, the valuewas slightly increased to 68% in

Mn-Biofilter and 95% in Zeo-Biofilter. On the other hand, the

TN removal in both biofilters was dropped down after WWTP

effluent plus Mn was applied. The median C/C0 ratios of TN

were increased from 86% to 93% in Mn-Biofilter and from 73%

to 93% in Zeo-Biofilter. The median C/C0 value of TN was

decreased to ~86% in both biofilters when Mn was excluded.

The removal efficiencies of UV254 and TP are present in

Fig. 6. Zeo-Biofilter showed a moderate removal of TP with a

median value of 41% which was much higher than the

removal in Mn-Biofilter (23%). However, UV254 was more

effectively removed in Mn- than Zeo-Biofilter: 51% vs. 22% of

median value. UV254 in the original WWTP effluent was

approximately 0.3 cm�1 and median values of UV254 were 0.15

Fig. 6 e Removal of UV254 and total phosphorus (TP) in Mn-

Biofilter and Zeo-Biofilter fed with WWTP effluent. The

whiskers show 5% and 95% percentiles and the solid dots

stand for the mean value.

Fig. 7 e Mn concentrations in the effluents of Mn-Biofilter

(a) and Zeo-Biofilter (b). The whiskers in boxplot show 5%

and 95% percentiles and the solid dots stand for the mean

value.

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6 223

and 0.24 cm�1 in the effluents ofMn-Biofilter and Zeo-Biofilter,

respectively. A significant influence of Mn in the feed was not

found on the removal of UV254 and TP in both biofilters.

3.3. Residual manganese

Manganese is normally not controlled for the effluent

discharge. WHO suggests a provisional value of 0.5 mg L�1

manganese for the hazardous control of drinking water while

a ten-fold lower value (50 mg L�1) is recommended by USA and

Europe to avoid odor and staining effects of manganese.

The spiked manganese was effectively removed with both

biofilters and its concentration in effluentwas generally below

the ahead mentioned standards in all operation stages as

shown in Fig. 7. However, a slightly elevatedMn concentration

was observed in the effluents of both biofilters when the real

WWTP effluent was applied to replace the synthetic waste-

water.Without spikingMn in the real wastewater, only a trace

amount of Mn was detected in both biofilters, which was in

the range of Mn content of applied WWTP effluent. It also

indicated that Mn was not released from the previously fixed

Mn on the packing materials.

4. Discussion

Current WWTPs cannot effectively remove SMX and DFC

(Larcher and Yargeau, 2011; Zhang et al., 2008b). Our results

provide an alternative solution to remove both substances

with a simple reactor and an ease operation. However, adap-

tation is required for both biofilters to effectively remove SMX:

1 month in Mn-Biofilter, 1.5 month in Zeo-Biofilter. It might be

related to the development of some SMX-degrading IMOB

since the removal of SMX should be a result of biological

manganese oxidation as shown in the control test. It is highly

possible that the Mn3þ intermediate formed during the bio-

logical oxidation of Mn2þ was involved in the degradation of

SMX, like the degradation of DFC by P. putida (Meerburg et al.,

2012). Nevertheless, a high level of SMX (2 mg L�1) was used

during the adaptation in this study. Baumgarten et al. (2011)

found that the high concentration of SMX can largely

shorten the adaptation time for SMX removal in aerobic sand

filters (HRT ~14 d). Therefore, it might be possible that both

biofilters in this study would need a longer adaption time

when starting immediately with secondary effluent where the

concentration of SMX is in the range of low mg L�1.

It took about 1 year to obtain an effective removal of DFC in

Zeo-Biofilter. It is interesting to notice that the removal of DFC

in Zeo-Biofilter was gradually increased along the operation

time, during which the spiked manganese should be biologi-

cally fixed in the form of MnOx on the zeolite (estimated as

~1.5 mg Mn per g zeolite according to the mass balance). After

the whole operation period, one can clearly found the black

biofilm on zeolite (Fig. 2s in Supplementary Information). A lot

of manganese was also detected on zeolite in the elementary

analysis with energy dispersive X-ray spectroscopy (Fig. 3s in

Supplementary Information). This Bio-MnOx on zeolite can

still eliminate DFC even with the presence of azide, however,

with a lower kinetic rate thanwithout azide as shown in Fig. 4.

It indicates that the involvement of biological Mn oxidation is

beneficial for the activity of Bio-MnOx, which is in agreement

with the study of Forrez et al. (2010).

On the other hand, Mn-Biofilter can achieve an immediate

removal of DFC, which should be attributed to the catalytic

effects of mineral MnOx, as demonstrated by the recent study

of Huguet et al. (2013). They also found that the oxidation

capacity of MnOx was gradually diminished during the reac-

tion with DFC (~20 mg L�1 in clean water). After the treatment

of 350 empty-bed volumes, the oxidation capacity dis-

appeared, which the authors attributed to the reductive

dissolution and saturation of the active sites. In this study, the

biofilters were fed with approximately 400 empty-bed

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6224

volumes of secondary effluent and a significant reduction of

DFC removal was not observed. The bioactivity in biofilters

may regenerate the active sites of MnOx or Bio-MnOx via e.g.

oxidizing the dissolute Mn2þ back to MnOx, which should be

confirmed with a separate study. Furthermore, it is worth

pointing out that the activity of Bio-MnOx is related to its

poorly crystalline layer structure (e.g. hexagonal sheet sym-

metry) (Villalobos et al., 2003). The structure of Bio-MnOxmay

change during the reaction and consequently its activity could

decrease. Recently, Learman et al. (2011) analyzed the evolu-

tion of structure and activity of Bio-MnOx from Roseobacter sp.

AzwK-3b during the in vitro oxidation ofMn2þ. They found that

78% of Bio-MnOx hexagonal symmetry was gradually trans-

ferred to a pseudo-orthogonal symmetry in 5 d and the latter

almost lost its activity of Mn2þ oxidation. It indicates that a

continuous generation of Bio-MnOx may be required to sus-

tain the catalytic activity. However, the structure evolution of

MnOx may also occur during the reaction with organic pol-

lutants, which is still not reported according to our

knowledge.

UV254 indicates the color and aromatic contents (e.g. humic

substances) in wastewater, which may be problematic, e.g. by

forming disinfection byproducts in chlorination and causing

membrane fouling (Tripathi et al., 2011; Zheng et al., 2010).

Wert et al. (2009) also found a strong correlation between

UV254 removal and pharmaceutical elimination from tertiary

treated wastewater by O3 and O3/H2O2. In their study,

4e8 mg L�1 of ozone dosage was required to remove 50% of

UV254. However, UV254 cannot be effectively removed by most

biofilters. For example, the slow sand filters can only remove

5e6% of UV254 from a secondary effluent (Zheng et al., 2010).

Wang et al. (2008) combined an ozonation column and a bio-

filter with a clay media to treat a WWTP secondary effluent.

They obtained 60% and 15% removal of UV254 in ozonation

(8 mg L�1 dosage) and biofilter, respectively. Lee et al. (2012)

applied a MBR to treat a WWTP secondary effluent, which

was followed by an ozonation column and a biofilter packed

with anthracite coal at EBCT of 20min. They found that 56% of

UV254 in theMBR effluent was removedwith ozone at a dosage

of 8 mg L�1 (Lee et al., 2012). The subsequent biofilter removed

negligible UV254. In the present study, about 50% of UV254 was

removed in Mn-Biofilter without the complex operation. It

indicates that Mn-Biofilter might also be applied as a pre-

treatment for membrane separation and drinking water

chlorination.

The low removal of carbon contents in both biofilters is

in accordance with other biofilters for the treatment of

secondary effluent: e.g. 25% COD removal (Wang et al.,

2008), 20% of DOC removal of pre-ozonated effluent

(8 mg L�1 ozone dosage) (Lee et al., 2012). Both biofilters had

a sufficient oxygen supply (dissolved oxygen 3e5 mg L�1)

and thus the denitrification cannot significantly occur.

Therefore, the limited removal of TN was found in both

biofilters. Nevertheless, a higher removal of DOC can be

noticed in Mn-Biofilter than Zeo-Biofilter (Fig. 5). It may be

due to the oxidation of humics by MnOx. It has been found

that chemically synthesized MnOx can mediate the oxida-

tion of humic substances to generate low-molecular-weight

organic compounds, including pyruvate, acetone, formal-

dehyde, etc. (Sunda and Kieber, 1994). That also partially

explained the better removal of UV254 by Mn-Biofilter.

However, the presence of humic substances could inhibit

the oxidation of organic pollutants. Li et al. (2012) found that

humic acid can dramatically decrease the mineralization of

phenolic monomers (catechol and p-coumaric acid) by

chemical MnO2.

It is also worth mentioning that the material cost of MnOx

ore per volumetric unit is more expensive (3e5 times) than

granular activated carbon. Therefore, the initial investment

for a MnOx filter should also be higher. However, activated

carbon is usually suffered from the declined adsorption ca-

pacity and consequently decreased removal of organics. Mn-

Biofilter did not encounter this problem. It is also possible

that activated carbon can further remove recalcitrant organics

thanks to the biofilm (Reungoat et al., 2012). However, the

development of mature biofilm may takes a long term, up to

several years. The natural zeolite is a less expensive material

but it took Zeo-Biofilter a long adaption term to degradeDFC in

this study. Its adaption may be shorten, e.g. by increasing the

loading ofmanganese, which surely should be confirmedwith

another study.

5. Conclusions and outlook

The following conclusions can be made from this study:

1) SMX and DFC can be effectively removed in the aerated

biofilter packed with natural MnOx and zeolite. However,

Zeo-Biofilter required a long adaption time (~1 year) to

remove DFC. Mn-Biofilter and Zeo-Biofilter can also effec-

tively eliminate the aromatic contents of DFC and SMX,

respectively.

2) DFC was mainly removed by the oxidation of MnOx while

SMX during the biological oxidation of manganese. The

bioactivity was also beneficial for the removal of DFC by

MnOx.

3) Spiked Mn2þ can be effectively removed and therefore

environmental impact of manganese feeding is minimal.

4) Both biofilters resulted in a limited removal of DOC and TN.

However, Mn-Biofilter obtained ~50% removal of UV254

from the secondary effluent, indicating a high potential of

eliminating aromatic contents in wastewater.

Further studies are required to shorten the adaption time

of Zeo-Biofilter, e.g. via increasing the loading of manganese

and to identify the crystal structure of Bio-MnOx and its in-

fluence on the elimination of DFC.

Acknowledgment

This study is part of the project TransRisk (grant No.

02WRS1276J) financed by the German BMBF within the

program of RiskWa. We also appreciate the work of Dr.

Wolfram Seitz, Zweckverband Landeswasserversorgung at

Langenau and Ms. Elena Kaiser, Bundesanstalt fur Gew€as-

serkunde (BfG) at Koblenz, for the analysis of

pharmaceuticals.

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6 225

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.watres.2015.01.009.

r e f e r e n c e s

Atlas, R.M., 1997. Handbook of Microbiological Media. CRC Press.Baquero, F., Martınez, J.-L., Cant�on, R., 2008. Antibiotics and

antibiotic resistance in water environments. Curr. Opin.Biotechnol. 19 (3), 260e265.

Basile, T., Petrella, A., Petrella, M., Boghetich, G., Petruzzelli, V.,Colasuonno, S., Petruzzelli, D., 2011. Review of endocrine-disrupting-compound removal technologies in water andwastewater treatment plants: an EU perspective. Ind. Eng.Chem. Res. 50 (14), 8389e8401.

Baumgarten, B., J€ahrig, J., Reemtsma, T., Jekel, M., 2011. Long termlaboratory column experiments to simulate bank filtration:factors controlling removal of sulfamethoxazole. Water Res.45 (1), 211e220.

Brausch, J., Connors, K., Brooks, B., Rand, G., 2012. In:Whitacre, D.M. (Ed.), Reviews of Environmental Contaminationand Toxicology, vol. 218. Springer, US, pp. 1e99.

Cerrato, J.M., Falkinham Iii, J.O., Dietrich, A.M., Knocke, W.R.,McKinney, C.W., Pruden, A., 2010. Manganese-oxidizing and-reducingmicroorganisms isolated from biofilms in chlorinateddrinking water systems. Water Res. 44 (13), 3935e3945.

De Rudder, J., de Wiele, T.V., Dhooge, W., Comhaire, F.,Verstraete, W., 2004. Advanced water treatment withmanganese oxide for the removal of 17a-ethynylestradiol(EE2). Water Res. 38 (1), 184e192.

Emerson, D., Fleming, E.J., McBeth, J.M., 2010. Iron-oxidizingbacteria: an environmental and genomic perspective. Annu.Rev. Microbiol. 64, 561e583.

Forrez, I., Carballa, M., Fink, G., Wick, A., Hennebel, T.,Vanhaecke, L., Ternes, T., Boon, N., Verstraete, W., 2011.Biogenic metals for the oxidative and reductive removal ofpharmaceuticals, biocides and iodinated contrast media in apolishing membrane bioreactor. Water Res. 45 (4), 1763e1773.

Forrez, I., Carballa, M., Noppe, H., De Brabander, H., Boon, N.,Verstraete, W., 2009. Influence of manganese and ammoniumoxidation on the removal of 17a-ethinylestradiol (EE2). WaterRes. 43 (1), 77e86.

Forrez, I., Carballa, M., Verbeken, K., Vanhaecke, L.,Schlusener, M., Ternes, T., Boon, N., Verstraete, W., 2010.Diclofenac oxidation by biogenic manganese oxides. Environ.Sci. Technol. 44 (9), 3449e3454.

Gouzinis, A., Kosmidis, N., Vayenas, D.V., Lyberatos, G., 1998.Removal of Mn and simultaneous removal of NH3, Fe and Mnfrom potable water using a trickling filter. Water Res. 32 (8),2442e2450.

Green, R.E., Newton, I., Shultz, S., Cunningham, A.A., Gilbert, M.,Pain, D.J., Prakash, V., 2004. Diclofenac poisoning as a cause ofvulture population declines across the Indian subcontinent. J.Appl. Ecol. 41 (5), 793e800.

Hennebel, T., De Gusseme, B., Boon, N., Verstraete, W., 2009.Biogenic metals in advanced water treatment. TrendsBiotechnol. 27 (2), 90e98.

Huguet, M., Deborde, M., Papot, S., Gallard, H., 2013. Oxidativedecarboxylation of diclofenac by manganese oxide bed filter.Water Res. 47 (14), 5400e5408.

Larcher, S., Yargeau, V., 2011. Biodegradation ofsulfamethoxazole by individual and mixed bacteria. Appl.Microbiol. Biotechnol. 91 (1), 211e218.

Learman, D.R., Wankel, S.D., Webb, S.M., Martinez, N.,Madden, A.S., Hansel, C.M., 2011. Coupled bioticeabiotic Mn(II)oxidation pathway mediates the formation and structuralevolution of biogenic Mn oxides. Geochim. Cosmochim. Acta75 (20), 6048e6063.

Lee, C.O., Howe, K.J., Thomson, B.M., 2012. Ozone and biofiltrationas an alternative to reverse osmosis for removing PPCPs andmicropollutants from treated wastewater. Water Res. 46 (4),1005e1014.

Li, C., Zhang, B., Ertunc, T., Schaeffer, A., Ji, R., 2012. Birnessite-induced binding of phenolic monomers to soil humicsubstances and nature of the bound residues. Environ. Sci.Technol. 46 (16), 8843e8850.

Meerburg, F., Hennebel, T., Vanhaecke, L., Verstraete, W.,Boon, N., 2012. Diclofenac and 2-anilinophenylacetatedegradation by combined activity of biogenic manganeseoxides and silver. Microb. Biotechnol. 5 (3), 388e395.

Murray, K.J., Tebo, B.M., 2007. Cr(III) is indirectly oxidized by theMn(II)-oxidizing bacterium Bacillus sp. strain SG-1. Environ.Sci. Technol. 41 (2), 528e533.

OECD, 2001. OECD Guideline for the Testing of Chemicals. Section3, Test No. 303.

Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F.,Keller, J., 2010. Removal of micropollutants and reduction ofbiological activity in a full scale reclamation plant usingozonation and activated carbon filtration. Water Res. 44 (2),625e637.

Reungoat, J., Escher, B.I., Macova, M., Argaud, F.X., Gernjak, W.,Keller, J., 2012. Ozonation and biological activated carbonfiltration of wastewater treatment plant effluents. Water Res.46 (3), 863e872.

Sabirova, J.S., Cloetens, L.F.F., Vanhaecke, L., Forrez, I.,Verstraete, W., Boon, N., 2008. Manganese-oxidizing bacteriamediate the degradation of 17-ethinylestradiol. Microb.Biotechnol. 1 (6), 507e512.

Sunda, W.G., Kieber, D.J., 1994. Oxidation of humic substances bymanganese oxides yields low-molecular-weight organicsubstrates. Nature 367 (6458), 62e64.

Tripathi, S., Pathak, V., Tripathi, D.M., Tripathi, B.D., 2011.Application of ozone based treatments of secondary effluents.Bioresour. Technol. 102 (3), 2481e2486.

Villalobos, M., Toner, B., Bargar, J., Sposito, G., 2003.Characterization of the manganese oxide produced byPseudomonas putida strain MnB1. Geochim. Cosmochim. Acta67 (14), 2649e2662.

Wang, S., Ma, J., Liu, B., Jiang, Y., Zhang, H., 2008. Degradationcharacteristics of secondary effluent of domestic wastewaterby combined process of ozonation and biofiltration. J. Hazard.Mater. 150 (1), 109e114.

Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Using ultravioletabsorbance and color to assess pharmaceutical oxidationduring ozonation of wastewater. Environ. Sci. Technol. 43 (13),4858e4863.

Zaman, M.I., Mustafa, S., Khan, S., Xing, B., 2009. Heavy metaldesorption kinetic as affected by of anions complexation ontomanganese dioxide surfaces. Chemosphere 77 (6), 747e755.

Zhang, H., Chen, W.-R., Huang, C.-H., 2008a. Kinetic modeling ofoxidation of antibacterial agents by manganese oxide.Environ. Sci. Technol. 42 (15), 5548e5554.

Zhang, Y., Geißen, S.-U., 2010. Prediction of carbamazepine insewage treatment plant effluents and its implications forcontrol strategies of pharmaceutical aquatic contamination.Chemosphere 80 (11), 1345e1352.

Zhang, Y., Geißen, S.-U., Gal, C., 2008b. Carbamazepine anddiclofenac: removal in wastewater treatment plants andoccurrence in water bodies. Chemosphere 73 (8), 1151e1161.

Zheng, X., Ernst, M., Jekel, M., 2010. Pilot-scale investigation onthe removal of organic foulants in secondary effluent by slow

wat e r r e s e a r c h 7 2 ( 2 0 1 5 ) 2 1 8e2 2 6226

sand filtration prior to ultrafiltration. Water Res. 44 (10),3203e3213.

Zhu, H., Sun, W., Zhang, Y., Szewzyk, U., Geissen, S.-U., 2012.Removal of pharmaceutical compound diclofenac by ironbacteria. In: Proceedings of the Conference of the Associationfor General and Applied Microbiology (VAAM), Tubingen,Germany.

Zimmermann, S.G., Wittenwiler, M., Hollender, J., Krauss, M.,Ort, C., Siegrist, H., von Gunten, U., 2011. Kineticassessment and modeling of an ozonation step for full-scale municipal wastewater treatment: micropollutantoxidation, by-product formation and disinfection. WaterRes. 45 (2), 605e617.